The paleontologist David Raup introduced the concept of a morphospace, the theoretical limits on morphological variation among organisms. Organisms past and present rarely occupy all areas of their theoretical morphospace, but we rarely know why. Stomata, the mouths of plants, are among the most important innovations for multicellular plant life on land. The size, density, and patterning of stomata on a leaf form a low-dimensional, mathematically tractable morphospace that applies to most land plants. Major regions of this stomatal morphospace are empty or sparsely occupied. Why? One hypothesis is that leaf development constrains the production of certain phenotypes. An alternative hypothesis states that the adaptive landscape itself is constrained and certain areas of morphospace are empty because they are never favored by natural selection. Comparative and genetic observations show that empty or sparse regions of stomatal morphospace are developmentally accessible. Furthermore the distribution of phenotypes is consistent with a model in which the adaptive landscape prevents further expansion into empty regions of morphospace. In the future, I would like to see whether optimality models can unite macro- and microevolution of stomatal traits by predicting limits on the adaptive landscape and the direction of natural selection under realistic conditions.